Forget Everything You Believed About Computing w/ Gill Verdon | EP #102
Added: May 24, 2024
In this podcast episode, Peter Diamandis interviews Gill Verdon, the founder of the Effective Accelerationism movement and CEO of Extropy AI. Verdon, a quantum physicist and visionary in AI and thermodynamic computing, discusses his ambitious goals to increase intelligence per watt and propagate intelligence throughout the universe. His goal is to create a brain-scale processor that is far more energy-efficient than the human brain, envisioning AI computing that is hundreds of thousands of times more efficient than current systems.Key takeaways
Gill Verdon's Effective Accelerationism movement emphasizes the importance of embracing technological change and maintaining variance to drive progress. This approach aims to adapt to challenges and foster a better future through technological advancements.
Verdon's goal is to create a brain-scale processor that is significantly more energy-efficient than the human brain. He envisions AI computing that is hundreds of thousands of times more efficient than current systems, leveraging thermodynamic and probabilistic computing principles.
Verdon introduces the Kardashev scale to measure civilization's progress based on energy consumption. He stresses the importance of aiming for higher goals, such as expanding into space, to continue humanity's growth and evolution.
Verdon advocates for decentralized models of AI augmentation to preserve individuality and foster innovation. He believes that centralization and control of AI pose greater risks than allowing individuals to access and control their own neural augmentations.
Verdon discusses the advantages of thermodynamic computing, which incorporates the inherent noise and heat of matter into the algorithm and hardware models. This approach offers significant benefits over traditional transistor-based computing, particularly in accurately modeling complex systems like chemistry and biology.
Effective Accelerationism: Embracing Progress and Adaptation
Verdon introduces the Effective Accelerationism movement, which focuses on understanding progress and advancement in civilization from a physical standpoint. This movement emphasizes the importance of maintaining variance and dynamism to adapt to challenges and drive progress towards better futures. He believes that embracing technological change and an abundance mindset is crucial for navigating rapid advancements in AI and computing.
Fostering Innovation Through Decentralized AI Augmentation
The conversation addresses the fear and polarization surrounding AI technology. Verdon acknowledges concerns about the power and potential risks of advanced AI but argues that centralization and control may pose greater dangers than giving individuals access to neural augmentations. He advocates for guiding the acceleration of technology towards a positive future and believes decentralized models of AI augmentation can preserve individuality and foster innovation.
AI Integration for Enhanced Intelligence
Verdon envisions integrating AI devices into human bodies within a certain thermal budget, allowing for more performant neural information processing. He foresees a world where AI augmentations are cheap, ubiquitous, and controlled by individuals, enabling them to harness cognitive leverage and augment their intelligence. He emphasizes the importance of maintaining individuality and variance in AI models to unlock the benefits of diverse perspectives and creativity.
The Role of AI in Accelerating Progress
The podcast explores AI as a powerful tool for accelerating progress and addressing societal challenges. Verdon highlights the role of the market in guiding the evolution of AI models, with users influencing the development of products through their choices and usage. He argues that centralization of AI control may hinder innovation and individuality, leading to a gradient of power and potential risks.
The Kardashev Scale and Humanity's Potential
Verdon introduces the Kardashev scale, a system that measures the progress of civilizations based on their energy consumption. He explains the different types of civilizations, ranging from Type 1, which harnesses the energy of its planet, to Type 3, which harnesses the energy of its entire galaxy. Verdon emphasizes the need for humanity to aim for higher goals and challenges, such as expanding into space, to continue growing and evolving.
Embracing Exponential Progress
The conversation delves into the concept of exponential progress and the importance of embracing technological change. Verdon encourages entrepreneurs to focus on deep tech and tackle hard problems in the physical world, as these areas offer more opportunities for innovation and growth. He highlights the significance of unique data sets and contrarian ideas in driving entrepreneurial success in the future.
Verdon's Quantum Computing Journey
Verdon shares his journey from studying quantum physics to working on quantum deep learning and quantum computing at Google. He discusses his transition to working on quantum communications and sensing, as well as his collaboration with Sergey Brin on quantum technologies and AI. Verdon's work focuses on building a full stack of hardware and software to create a new branch of computing that combines thermodynamic and probabilistic principles.
Efficiency through probabilistic computing
The podcast explores the concept of thermodynamic and probabilistic computing, where computers operate in a state of uncertainty rather than strict determinism. Verdon references Maxwell's demon thought experiment to explain the energetic cost of knowledge and how running AI algorithms on deterministic digital computers is inefficient. By running probabilistic programs on probabilistic computers, there is higher efficiency and tighter integration of algorithms with hardware.
Thermodynamic Computing Advantages
Verdon explains that traditional transistor-based computing faces limitations as transistors become smaller and more energy-efficient due to the inherent noise and heat generated by the jiggling of matter. Instead of trying to eliminate this noise, thermodynamic computing incorporates it into the algorithm and hardware models, making it an integral part of the system. He emphasizes that this approach is a necessary evolutionary step in computing and offers significant advantages over traditional methods.
Accurate Modeling of Complex Systems
One of the key benefits of thermodynamic computing is its ability to accurately model complex systems such as chemistry and biology. Verdon highlights that traditional computing struggles to simulate the dynamics of proteins and molecular interactions due to the rapid fluctuations at the mesoscale level. By embedding the physics of protein folding and molecular dynamics into the hardware, thermodynamic computing offers a more efficient and accurate way to simulate these complex systems.
From Theoretical Physicist to Thermodynamic Computing Innovator
Verdon shares his journey of developing the concept of thermodynamic computing, starting from his early days as a theoretical physicist to his time at Google, where he gained valuable experience in quantum machine learning. He describes how he came up with the idea for thermodynamic computing and the ambitious goal of maximizing the amount of intelligence in the universe while optimizing intelligence per watt.
Revolutionizing AI with Thermodynamic Computing
The conversation delves into the potential applications of thermodynamic computing, particularly in the field of artificial intelligence. Verdon envisions a future where thermodynamic chips power AI systems at both the edge and in the cloud, offering unparalleled speed and energy efficiency. He emphasizes the importance of open-sourcing the concept of thermodynamic computing to allow a broader community to contribute to its development and adoption.
Advancements in Thermodynamic Computing
Verdon discusses the technical aspects of thermodynamic computing, including the embedding of algorithms into the physics of the hardware. He explains how thermodynamic programs can be compiled into primitives that align with the physics of the device, resulting in faster and more energy-efficient computations. He envisions a future where thermodynamic chips outperform traditional semiconductor chips, leading to widespread adoption of intelligent systems embedded in various devices.
Rapid Technological Advancement and Adaptability
The conversation also touches on the rapid pace of technological advancement and the need for individuals and organizations to adapt to the changing landscape. Verdon emphasizes the importance of maintaining an open mind and embracing variance and adaptability to navigate the uncertainties of the future. He invites engineers and AI experts who are looking for a new challenge to join his team at Extropy AI and contribute to pioneering the field of thermodynamic computing.
Related episodes
